GitLab

2 columns instead of 3; also improved style a bit.
Still a ways to go.

Note the following: (from the API file)
  IMPORTANT: elvin uses timeouts internally. So, this function does
  NOT guarantee that when it returns, either an event has been
  recieved or your timeout has passed. This should not be much of
  a problem, but you have been warned!

The above is not really fixable, without hacking elvin. And it may
not be entirely fixable even then. In particular, the first call to
event_poll_blocking() will always return at once, since there are
leftover timers from connecting to elvind.

Added "thumbnail view" to Experiment List page.
Added thumbnail rendering to renderer.

Note that thumbnail view is not available when viewing the idle list.

Also, loading thumbnails for "all" as admin takes a while!

for lan node delays). Also, return the vnode names for both delays
and linkdelays so that the agent knows what vnode it is operating on
behalf of. This is for the event subscription. The intent is ti get
away from the pipe number stuff.

params to the entry, so that two pipes are built, one to the lan
and one from the lan. For duplex links, these params are ignored since
there is only one pipe (to the switch).

New visualizer; uses "vis_nodes" table in database to store precomputed solutions.
Added "prerender" call to tbprerun and tbend.
"prerender_all" can be used to precompute vis solutions for all experiments.
Also, removed legacy vis cruft.

message screws up the redirect!

Note: Need to remove the silly "please be patient" messages. I've been
removing those and adding redirects since these messges are never seen
(the page is not drawn until the closing </table> is sent over). I
have a little javascript thing that throws up a be patient box, that
disappears when the page is finished. Need to get that installed some
day. Worked fine on opera and explorer, but on netscape it looked
totally stupid. Netscape, what else is new.

  leader wants to be in the subgroup, he has to do it via the editgroup
  page. This required minor changes in editgroup pages, since I was special
  casing the project leader to not allow removal/addition.

* Allow mere users to be the head of a group. This was previously not
  allowed, and is totally wrong since the entire group trust mechanism
  is based on giving subgroup members *more* privs then they have in
  the default (project) group.

* Change permission check in the showgroup page to allow non group members
  to look at the group if they have group_root or better in the default
  group. I noticed that once I took myself out of a group, I could no longer
  look at the group even though I had group_root in the project.

  Also change so that the edit/del menu does not appear unless the user
  has permission to do those things.

* Change consistency check when adding a group member. New test is simpler
  and makes sure that the user does not have root privs in the project and
  user privs in the subgroup. The reverse is of course okay, and the expected
  manner in which groups should be used.

* newgroup page now spits out a redirect to showgroup page, rather then
  printing the group info itself. Avoids duplication and gets rid of the
  form post from the history. Ditto for editgroup page.

scripts.  Provides some fancy output, methods to create and edit
files, and ways to check to see if something's already been done. Also
has some facilities to help debug why it's failed, without cluttering
normal output.  Should help make the installtion scripts more robust.

Contains a few tricks that are likely to get me my very own circle
in Perl Hell.

treatment. Fix up the scripts that use them, so that they get
quoted correctly.

before allowing a project to be approved. (In approveproject_form,
this meant removing the automatic transition from 'newuser' to
'unverified')

Since this check is done in two places, once in the form, and once in
the page that does the work, it's possible for this to get out of
sync.

picking users from projects with swapped out experiments!

Fixed problem with idle experiment table (it was due to an unclosed <p> tag.) While I was here, did a touch of sprucing.

link delays setup on the nodes.

Add tb-set-forcelinkdelays to force a linkdelay (when uselinkdelays is
on of course) to be inserted, even if the link was unshaped. This
allows it to be controlled later, say via the event system. This is
nice, cause in the current system an unshaped link cannot be later
shaped.

Note that these are currently require FBSD-47.

previously to perl generated email).

	when loading a slice, set the slice type in the DOS parttable
	run ntpdate if available, and echo the date

loaded on at once. The idea is to use this for things like rtLinux
that require a per-machine license. This mechanism is circumventable,
(for example, by simply making a second image), so it's mainly to
prevent experimenters from accidentally violating the license.

condition between two snmpits.

file with a tb- command and used in assign_wrapper to *force* link
delays when a delay would not otherwise be inserted. This avoids the
problem of trying to figure out when the user has set a queue param
that requires a delay. Kludgy, I know.

file with a tb- command and used in assign_wrapper to use link delays
instead of delay nodes. Default is zero, of course.

tweaks.
Add IPALIASES to ifconfig when (vers > = 8). This is the initial
approach for doing emulated links.

(frozen) or for webonly users.

determine if the user actually changed anything that requires a
callout to the scripts, and to determine if email needs to be sent.
Remove change password indicator since the audit list sees that anyway
via new "tbacct passwd" command. Change BCC/CC on the email. The
informative email goes to the user only now (audit list gets the
script mail), but CC audit whenever the email address is changed.